Re: [patch, mips] Improved memset for MIPS

["Joseph S. Myers" <joseph at codesourcery dot com>]

On Fri, 6 Sep 2013, Steve Ellcey wrote:

> On Fri, 2013-09-06 at 14:30 +0000, Joseph S. Myers wrote:
> > On Thu, 5 Sep 2013, Steve Ellcey  wrote:
> > 
> > > Tested with the glibc and gcc testsuites and by doing some standalone
> > > performance measurements.
> > 
> > Has the glibc testsuite been run without regressions for all six 
> > combinations of (o32, n32, n64) with (big-endian, little-endian)?
> 
> No.  I did most of my testing outside of the glibc testsuite because I
> find the glibc testsuite difficult to run, see
> https://sourceware.org/ml/libc-help/2013-08/msg00040.html for some of my
> problems/questions.  I don't believe I have ever managed to do a clean

You'll need to debug the problems as they indicate something wrong with 
your build environment.  It's always advised to configure glibc with 
--prefix=/usr rather than some other prefix (but there is no requirement 
that the dynamic linker actually be installed during testing, you can 
ignore the -dynamic-linker= path), and your other error indicates some 
inconsistency regarding NO_CTORS_DTORS_SECTIONS.

If you see more failures than are described at 
<https://sourceware.org/glibc/wiki/Release/2.18>, you should investigate 
them as well.

The expectation is that the glibc testsuite is the normal way to test 
patches before submission, and string function patches like this need it 
to be run for all six relevant ABI variants.

-- 
Joseph S. Myers
joseph@codesourcery.com